Performance Comparison of Convolutional Neural Network Models for Plant Leaf Disease Classification

Abstract

As food is an essential element in life, modern possibilities must be harnessed to pay attention to it. In this paper, we will discuss the discovery of early plant diseases classification using artificial intelligence technology, we made in this study analysis of convolutional neural networks architecture (vgg-16, mobile net, efficient net) and made a comparison between these model in accuracy and loos data in each model, we used data set from the Kaggle site that contain 20640 picture from different disease of plant (potato, tomato and pepper) this pictures divided on 15 class but unbalanced. at the first we solved the problem of train the models with multi class, we made balanced data and training the work in environment of Google Colab, we used it in train three models vgg-16, mobile net, efficient net, which showed this study that the accuracy of work in efficient net is 98% more than other models and loss data in this model is less than other models.

Plant phenotyping relevance

植物葉の病害状態を画像から分類するCNN手法を比較・評価しており、病害表現型の抽出とモデル性能比較が研究の中心です。

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.